Hugging Face's logo

zuazo
/
whisper-large-v3-eu-cv16_1

Automatic Speech Recognition
Transformers
TensorBoard
Safetensors
Basque
whisper
whisper-event
Generated from Trainer
Eval Results
Inference Endpoints
Model card Files Metrics Community
whisper-large-v3-eu-cv16_1 / README.md
zuazo's picture
zuazo
End of training
0e62997 verified
preview code
|
raw
history blame contribute delete
4.38 kB
metadata
language:
  - eu
license: apache-2.0
base_model: openai/whisper-large-v3
tags:
  - whisper-event
  - generated_from_trainer
datasets:
  - mozilla-foundation/common_voice_16_1
metrics:
  - wer
model-index:
  - name: Whisper Large-V3 Basque
    results:
      - task:
          name: Automatic Speech Recognition
          type: automatic-speech-recognition
        dataset:
          name: mozilla-foundation/common_voice_16_1 eu
          type: mozilla-foundation/common_voice_16_1
          config: eu
          split: test
          args: eu
        metrics:
          - name: Wer
            type: wer
            value: 6.887994372362044

Whisper Large-V3 Basque

This model is a fine-tuned version of openai/whisper-large-v3 on the mozilla-foundation/common_voice_16_1 eu dataset. It achieves the following results on the evaluation set:

  • Loss: 0.3688
  • Wer: 6.8880

Model description

More information needed

Intended uses & limitations

More information needed

Training and evaluation data

More information needed

Training procedure

Training hyperparameters

The following hyperparameters were used during training:

  • learning_rate: 1e-05
  • train_batch_size: 32
  • eval_batch_size: 16
  • seed: 42
  • gradient_accumulation_steps: 8
  • total_train_batch_size: 256
  • optimizer: Adam with betas=(0.9,0.999) and epsilon=1e-08
  • lr_scheduler_type: linear
  • lr_scheduler_warmup_steps: 500
  • training_steps: 40000
  • mixed_precision_training: Native AMP

Training results

Training Loss Epoch Step Validation Loss Wer
0.0095 10.04 1000 0.2023 9.6803
0.0032 20.08 2000 0.2153 9.0521
0.0023 30.11 3000 0.2234 8.8645
0.0023 40.15 4000 0.2278 8.4366
0.0012 50.19 5000 0.2260 7.9911
0.0005 60.23 6000 0.2435 7.9060
0.0013 70.26 7000 0.2254 7.8484
0.0004 80.3 8000 0.2367 7.4830
0.0008 90.34 9000 0.2289 7.4420
0.0007 100.38 10000 0.2385 7.5319
0.001 110.41 11000 0.2293 7.6325
0.0001 120.45 12000 0.2473 7.1430
0.0001 130.49 13000 0.2488 7.1870
0.0004 140.53 14000 0.2398 7.1831
0.0 150.56 15000 0.2620 7.0590
0.0001 160.6 16000 0.2547 7.1967
0.0 170.64 17000 0.2768 7.0736
0.0 180.68 18000 0.2878 7.0004
0.0 190.72 19000 0.2962 6.9466
0.0013 200.75 20000 0.2354 7.6042
0.0 210.79 21000 0.2720 6.8948
0.0 220.83 22000 0.2865 6.8987
0.0 230.87 23000 0.2954 6.8890
0.0 240.9 24000 0.3031 6.8821
0.0 250.94 25000 0.3102 6.8772
0.0 260.98 26000 0.3166 6.8899
0.0 271.02 27000 0.3233 6.8919
0.0 281.05 28000 0.3248 6.8919
0.0 291.09 29000 0.3363 6.9026
0.0 301.13 30000 0.3419 6.9085
0.0 311.17 31000 0.3471 6.8851
0.0 321.2 32000 0.3526 6.8704
0.0 331.24 33000 0.3570 6.8831
0.0 341.28 34000 0.3614 6.8851
0.0 351.32 35000 0.3645 6.8782
0.0 361.36 36000 0.3663 6.8714
0.0 371.39 37000 0.3677 6.8675
0.0 381.43 38000 0.3681 6.8802
0.0 391.47 39000 0.3686 6.8880
0.0 401.51 40000 0.3688 6.8880

Framework versions

  • Transformers 4.37.2
  • Pytorch 2.2.0+cu121
  • Datasets 2.16.1
  • Tokenizers 0.15.1